This paper describes a new tool developed for the detection of operating faults in ventilation units with heat recovery. In principle, the tool is based on the APAR (Air Handling Unit Performance Assessment Rules) method. By following the semantic data description in accordance with the BrickSchema and Project Haystack initiatives, the tool is portable. The executive part of the fault detection system consists of several dozen detection rules, which simultaneously seeks to estimate wasted energy, the threat to user comfort, or the risk of reduced device lifespan, so that the detected faults can be sorted according to their severity. The developed detection tool was validated on real devices incorporated in a pilot plant. For validation purposes, the method of fault induction on real HVAC (Heating, Ventilation and Air Conditioning system) units was used, with subsequent inspection of whether the faults were revealed or not. The results revealed a 90% detection rate. The data set created as a result of this pilot plant is published as an annex to this article. In addition, the ability of the detection tool to reveal faults was also verified on the basis of data sets of measurements taken during the standard operation of several dozen HVAC units. The elimination of the identified operating faults generated energy savings of several thousands of dollars per year.